Reinforcement learning produces dominant strategies for the Iterated Prisoner’s Dilemma

   Abstract